Stress relaxation in the lamellar copolymer mesophase

Feb 1, 1990 - Yi Zheng, You-Yeon Won, Frank S. Bates, H. Ted Davis, and L. E. Scriven , Y. Talmon. The Journal of Physical Chemistry B 1999 103 (47), ...
1 downloads 0 Views 2MB Size
Macromolecules 1990,23, 824-829

824

Stress Relaxation in the Lamellar Copolymer Mesophase T.A. Witten,' L. Leibler,+and P. A. Pincus* Exxon Research and Engineering Company, Annandale, New Jersey 08801. Received February 24, 1989; Revised Manuscript Received J u n e 19, 1989 ABSTRACT: When diblock copolymers organize into a lamellar mesophase, the interpenetrationbetween like blocks from opposing interfaces is inhibited at high molecular weight. For layers of height h and polymerization index N we show that interpenetration is confined to distances [ of order (h/R")-1/3,which becomes much smaller than h itself. We show by self-consistentfield methods that the concentration of penetrating monomers falls off with penetration depth y as exp(-(~/[)~/~). We argue that the relaxation time for sliding stress grows exponentially as a fractional power of the molecular weight because of this small interpenetration. A numerical example calculation suggests that interpenetration is significantly inhibited in real copolymers and that the predicted slow-relaxation regime should be accessible.

I. Introduction Block copolymers made with incompatible blocks form equilibrium domain structures of various types.' These structures and their distinctive properties have attracted steadily increasing interest.' These copolymers have great practical utility in improving properties of homopolymer blends3 and of adhesive^.^ Thus it is important to understand how the rheology of these materials differs from that of simpler polymer fluids. Empiri~ally,~ this rheology is marked by high viscosity, leading to difficulty in processing. It is also observed that the domains may be aligned by shear; the domain walls tend to line up along the stream lines.6 Our aim here is to explore the distinctive rheological features expected for copolymer domain phases. Distinctive features are expected because of the unusual state of the olymers in the domains. These polymers are stretched'+ out to lengths indefinitely greater than their ideal random-walk dimensions. Moreover the junction points in these copolymers are confined to the domain interfaces, thus strongly limiting their motion. Rheological behavior in weak flow may be described in terms of the stress response to a small step strain. This response is characterized by the initial stress and by the relaxation time required for this stress to decay. We consider two cases: a shear tending to slide the layers past one another and a compression of the layers. As anticipated, we calculate a distinctive behavior of the relaxation time for these two types of stress. A common feature of the two is an exponential growth of time with molecular weight, owing to the end confinement of the copolymer chains. More remarkable is the behavior of the sliding response time. This grows exponentially in a fractional power of the molecular weight and thus becomes much shorter than the compressional time. This relatively fast response arises from the distinctive small interpenetration of chains from opposing layers. We consider chiefly the case of a neat phase of symmetric diblock lamellae where both blocks are in the melt state. Since we are interested in scaling behavior at high molecular weight, it suffices to use schematic models of the polymer coils.'o Since the effects important here are at length scales longer than the distance between these coils, it suffices to treat their interactions with self-consistent field methods.'0*" The relaxation processes we Present address: Physico-Chimie Thbrique, ESPCI, 75231 Paris, France. Present address: Materials Department, Univ. of California, Santa Barbara, CA 93106. 0024-9297/90/2223-0824$02.50/0

treat arise from the topology and connectivity of the chains, so that the motions may be described at a schematic level." Our estimates deal with the asymptotic behavior for high molecular weight,' where the recent conformational theory of S e m e n ~ v is' ~applicable. ~~~ In order to discuss the stress relaxation we first describe the interpenetration zone, where chains from opposing interfaces may coexist. We then discuss the chain motions required for the relaxation of sliding stress and estimate the corresponding relaxation time. We next deal with the case of compressional stress. Finally, we estimate the various times and moduli for a specific example taken from recent structural studies.

11. Interpenetration Zone The copolymer blocks in a mesophase domain may be treated15 as statistically independent random walks in a nonuniform external potential. For our purpose we may treat each polymer as a chain of N beads connected by harmonic springs of zero unstretched length. In a lamella each bead has a height-dependent free energy denoted p ( z ) , measured in units of kBT. This p is the free energy cost of bringing that bead to the height z from some reference position. The probability G(r,r',12) that a subchain of 12 beads starting at r' ends at r obeys''

aG/ak = (1/2)a2V2G- p ( z ) G (1) where the microscopic length a' is related to the size of the free chain (without p): ((r - r')') = 3a'k. From this G the bead density p at any height z may be inferred." For diblocks in a lamella, each block, of N beads, has one end confined within the narrow phase boundary between the blocks. The other end is free. We denote the number of chains per unit area of the interface as u. The potential &) arises from the interaction of a given chain with others. In an incompressible melt p ( z ) must be such as to achieve a constant, predetermined density of beads p everywhere. Once this p is determined, the chain conformational statistics can be readily found from eq 1. As shown in ref 14 the problem of finding the potential profile p ( z ) simplifies greatly at high molecular weight N . Then each chain ending at zo may be regarded as a stretched-out line of beads which fluctuates only narrowly in position, as shown in Figure 1. Each bead is in mechanical equilibrium under the tensions from its successor and predecessor, together with the external force -Vp from the other chains. The only important randomness in the ensemble of chains is then the distribution 0 1990 American Chemical Society

Stress Relaxation in a Copolymer Mesophase 825

Macromolecules, Vol. 23,No. 3, 1990

Li

h

2.3

height z

Figure 1. Sketch of the asymptoti~'~ monomer free energy r(z) in a lamella of height h with schematic representation of a copolymer hloek of N heads with its free end at height zo. The numbering of the "beads" is shown.

energy. Such energies are capable of distorting the penetrating segment away from its random-walk configuration. For a penetrating section of k beads, the interaction energy cost of penetration is pk N h(z - h)k/(R2N. The stretch energy of such a section is roughly [(z - h)/ R]'(N/k). The sum of these is minimal when k2 N P ( z - h)hh. The net energy cost AS(z) is then roughly h1I2(zh)3/ /R2. The concentration of chain ends penetrating to a height z is [const X exp(-AS(z))], so that the numher of chain ends falls off exponentiallywith the 3/2 power of the penetration depth. We now corroborate these qualitative expectations using more systematic calculations. We consider first the penetration of homopolymers into the grafted layer. The local concentration $(z) of the penetrating chains may he found by using Edward's ground-state dominance Here, the concentration @ becomes the square of an amplitude $, which satisfies (l/Z)a2V2$

t(zO) of starting heights. The potential p ( z ) must he = A - Bz2-in this limit, for this is the parabolic-&)

only form that takes a chain starting from any height zo to the z = 0 interface in exactly N heads. The coefficient B is given hy B = a - 2 r Z / ( 8 P ) The . A coefficient depends on the coverage u: A = Bh2, so that p = 0 a t the height h that accommodates all Nu heads per unit area at the required head density p , viz., h = uN/p. In this large-N limit the grafted layer becomes impenetrable to outside chains. The reduction in penetration for increasing N has been observed numerically" and experimentally? Because our copolymers are stretched, it is rather difficult to find the amount of penetration analytically. To get some insight on the extent of the interpenetration zone we consider first the sample system of a grafted-chain layer in contact with a melt of long, free homopolymers. The penetration of homopolymer into the grafted layer is quite important in itself for understanding certain aspects of adhesion4 and compatibilization3of polymer blends. Having analyzed this simple penetration problem, we turn to our central question: viz., the interpenetration of chains across a lamella. Qualitatively, the extent of penetration hy an outside chain of length k is controlled by the competition between the translational entropy to he gained by penetration and the energetic cost of penetration." Thus, penetration into a region of potential p is inhibited whenever pk 2 1. For the majority of the layer, p N Bh2 N l/N(h/R)', where R = 3a2Nis the root-mean-square end-to-end distance of the unperturbed chain. Thus penetration throughout the layer is only possible for small chains of length k < N(R/h)'. Chains much longer than this k may only penetrate into the outer fringe of the copolymer layer, where the potential is much smaller than its maximum value: p N Bh(z - h) N h(z - h)/(R2N). A penetrating segment of length k may penetrate a distance z - h of order ak'l', since any potential that allows substantial penetration cannot distort the random-walk statistics drastically. Penetration into the outer fringe is allowed when pk N 1,as above. This leads to a condition for penetration k3I25 N3I2(R/h). The chains penetrate to distance (h - z) of order ak'I2, i.e. (h - z ) N~ R3(R/h) P / h . Even this fringe penetration is inhibited whenever the copolymers are stretched so that h > R. Of course, some outside chains penetrate further than the characteristic distance (R4/h)'I3. We may understand these deeply penetrating chains by the same scaling arguments used above. A chain penetrating far beyond the typical distance necessarily pays a high price in free

-

- pJ. = 0

(2)

with J" = @ = p outside the grafted layer. The constant a is the microscopic length introduced in eq 1. Evidently @ decays to zero within the layer with some characteristic length & For the purpose of determining the potential p this interpenetration may he neglected, so that we may use the asymptotic parabolic form of p ( z ) . Thus near the extremity of the layer, p ( z ) = (z - h)F, with F = a-'2a2h/(4P). In view of eq 2 one expects on dimensional grounds F2 FE, or 5 = P I 3 . More concretely, the assumed form r#~(z)= p exp(f([h - z]/s)) yields for h - z >> 5

-

+(z) = p exp[-(4/3)((h - Z ) ~ / ~ F ' / ~ ) ]

(3) up to a constant numerical factor. (For later use we note that the amplitude $ behaves the same, with the 4/3 replaced by a 2/3.) Since E P I 3 h-'I3Pl3, the interpenetration region grows more slowly than the height, as claimed. The thickness 5 agrees with our qualitative estimate for (z - h) above. If the coverage u is fixed, then h Nand E N113. If instead h PI3,as in block-copolymer lamellae: then E N4I9. The penetrating chains alter the potential p from its parabolic form-indeed" the potential is altered in this region by corrections to the asymptotic classical behavior, even without the homopolymer. Still, the potential is only slightly altered in the deep region described by eq 3, so that this equation remains valid. Since this limit of the penetration zone is correct, we expect the formula to he qualitatively correct for the hulk of the zone, so that our conclusions about scaling behavior remain valid. The same decrease in penetration occurs if the penetrating chains come from the opposing layer in a pure dihlock system. Interpenetration arises because a given end may move slightly above a height h without much cost in its (free) energy S(z). In the strongly stretched limit appropriate for long chains13*'" this energy has the form

- -

- -

S(z,) = x(1/2)a"(dz/dk)2

-

-

+ p(z(k))

k

where z(k) is the height of the kth head of the chain and a, defined above, is a microscopic length. Using the parabolic p ( z ) cited above s(~,) is independent of zo for all zo < h.'" But if zo is raised above h and into the opposing layer, S begins to increase. (The exact, self-consistent p ( z ) is not parabolic, but we assume for the moment that the deviations from the parabola may he neglect-

Macromolecules, Vol. 23, No. 3, 1990

826 Witten et al.

ed.) A straightforward classical-mechanics calculation, reported in the Appendix, gives AS = (2/3)[8F/3]'/%-'(h- z ) 3 / 2 (4) As noted above, we expect chain ends to extend above the height h if the energy cost AS is of order kT or less. From eq 4 this amounts to y ( z - h) N u2f3F1f3. In the parabolic potential the force F is a-27r2h/(4N2).Thus ends from one parabola are to be found in the opposing ' f ~y. parabola over distances y of order ( h ~ - ~ / P ) -This is the width of the interpenetration region. A chain beginning near the height h is not strongly stretched over the distance y. If the first k beads of a chain were unperturbed by stretching effects, their meansquared displacement ((6~)~) would be of order a'k. The stretching induced by the force F gives rise to a net displacement ( A z ) of order k dz/dk N (Fa2k2). Evidently the stretching contribution to Az becomes dominant for ( F u ~2 ~a2k, ~ i.e., ) ~for Az 2 (Fu)-'f3 ( h / P ) - 1 / 3This . Az is evidently on the order of y above. Thus the chain sections in the interpenetration region, unlike the chains as a whole, are only weakly stretched. Qualitatively, these sections may be treated as random walks; a chain in the interpenetration zone y has roughly (y/uI2 beads there, as anticipated above. The occurrence of interpenetration alters the potential p from the parabolic form assumed here. Thus, our estimates of the zone width y are not obviously self-consistent. The true self-consistent potential for two opposing grafted layers is softer than the two abutting parabolas assumed above. Still, the true potential can be perturbed but little at the extremes of the interpenetration zone, so that dp/dz at z = h - y must still be of the order F assumed above. Thus, our estimate of y should not be altered importantly by the corrections to the potential. However, such corrections would be necessary in calculating the shape of the interpenetration zone in any detail. The explicit scaling of the zone width y with chain length N depends on the case. For copolymer chains in a lamelas Pf3. Accordlar phase, the la er thickness h ~aries~9'~9'~ ingly, y N4f . For chains end grafted to a surface in a solvent, the height h depends on the number of chains per unit area u and on the strength of the excluded-volume interaction between the beads. For fixed coverage u the height h is proportional to N , so that y N1f3.

-

- r

-

111. Relaxation of Sliding Stress

The limited interpenetration of chains in opposing layers leads to limited entanglement. Thus we expect that stress held by these entanglements should relax in a distinctive way. For comparison we recall the behavior of a melt of homopolymer chains of length N . If a step strain is imposed on an entangled homopolymer melt at time t = 0, the resulting stress resembles that in a rubber network. The associated "plateau modulus" arises2' from the distortion of the mutually entangled chains in the melt. The stress relaxes from this rubberlike plateau in a time 7 that increases strongly with the chain length N . The viscosity of the melt is proportional to this T . The relaxation time is believed12 to be the time required for a typical chain to escape from its initial entanglements. The chains are believed to escape by "reptation", a Brownian motion of a chain along its contour length. Asymptotically, this escape requires a time T of order P. We may apply a similar step strain to a lamellar domain structure. We consider here a transverse shear strain tending to slide each lamella along its neighbors. For sim-

plicity we suppose that the two blocks are symmetric and that both are in a melt state above any glass transition temperature. As in the homopolymer melt, the initial shear distorts the entangled copolymer chains. The number of entanglements per unit volume is a local property of the melt; thus the associated modulus should be the same as in the homopolymer melt discussed above. But these chains cannot escape from their entanglements in the same way as homopolymer chains. On the other hand, they each have one end anchored a t the domain interface, so that reptation is impossible. On the other hand, they are stretched substantially beyond their equilibrium size. This could produce a driving force for relaxation. Further, as noted above, only a limited escape is required. Only chains entangled with opposing chains transmit stress between layers. Thus we assert that the time 7, for stress to relax is the time for a chain in the interpenetration zone to disengage from that zone. A chain end in the interpenetration zone may over time retract into its own layer. Indeed, chain ends are di~tributed'~ throughout the layer, so that a given end spends only a minority of its time in the interpenetration zone. As in an ordinary polymer melt, a chain may move along its own contour without violating entanglement constraints. This motion results from independent Brownian forces on the beads of the chain, i.e., Rouse dynamics.12 The motion may be described in terms of local elements of stored length-"reptons"-along the contour.21 These reptons diffuse along the chain independently. Their diffusivity depends only on the local structure of the chain and its environment; thus it is the same as in the homopolymer melt. Whenever a repton enters or leaves the end of a chain, that end advances or retracts a small distance. As with homopolymers, this local distance, or "tube diameter" is on the order of a monomer length. The curvilinear displacement of an end is subdiffusive at short times, as with homopolymers. Specifically, the root-mean-square curvilinear displacement k( t ) over time t varies as t ' / 4 for times t